Claude Fable 5.2: when to use it instead of Opus 5.5, Sonnet or Haiku

ASC's thesis is simple: pick the right model for each job. Anthropic's Nov 9, 2026 launch of Claude Fable 5.2 and Mythos 5.2 adds a top tier for the hardest technical work, while keeping Opus 5.5 as the recommended daily driver. Here is how we read the official guidance so you can match Fable, Opus, Sonnet and Haiku to the task in front of you.

The tools

Claude Code

Anthropic's coding agent where Fable 5.2, Opus 5.5, Sonnet and Haiku show up as model choices for long-running engineering work. This is the natural place to try Fable when a repo-scale migration or ambiguous system needs a stronger orchestrator.

Use it for: Ambitious coding sessions, migrations and multi-agent orchestration with Claude models.

Claude Pro+ / API

Claude

Claude on the web and apps, where paid plans can use Fable 5.2. Anthropic is giving eligible paid plans extra Fable usage in the launch week, which is a low-risk way to test when Fable beats Opus 5.5 for your workflows.

Use it for: Chat, research and knowledge work when you want Fable without wiring the API yourself.

Free / paid

Cursor

A vibe-coding IDE that can call frontier models for edits and agents. Pair it with Anthropic's guidance: keep a strong daily-driver model for most edits, and reserve Fable-class capability for open-ended, high-stakes builds.

Use it for: Day-to-day product coding where model choice still matters for hard problems.

Free / $20+ mo

Descriptions and prices come from each tool's official site, checked Nov 9, 2026. Prices change often, so confirm before you buy.

Practical tips

1. Start with Opus 5.5 as the daily driver

Anthropic's own guidance is to begin with Claude Opus 5.5 for most coding and knowledge tasks, then move up to Fable or Mythos only when the scope and complexity demand it, or down to Sonnet 5.5 or Haiku 5.5 when a lighter model is enough. That matches ASC's pick-the-right-tool-for-the-job habit.

2. Reserve Fable for ambitious, open-ended work

Fable 5.2 and Mythos 5.2 are aimed at the hardest technical challenges: large coding projects, scientific reasoning, hardware and robotics-style control, and long agent runs. Anthropic's showcases include a working OS built with 1,000+ subagents, Fermat's Last Theorem formalized in 30 hours, and a Python-to-Rust production migration.

3. Know the Fable vs Mythos split

Fable and Mythos are the same underlying model. Fable includes cybersecurity and biology safeguards so it can ship to all users. Mythos is for vetted organizations through Anthropic's Life Sciences and Cyber Verification Programs when those dual-use domains are part of the work.

4. Use Fable as advisor or orchestrator

Anthropic notes that Fable 5.2 can advise smaller models (Haiku, Sonnet or Opus consulting it on harder problems) or orchestrate them by delegating simpler tasks. That is a clean pattern for cost control: keep cheap models on the routine steps and call Fable when judgment or coordination matters.

5. Price and speed: standard vs fast mode

API pricing is $8 per million input tokens and $40 per million output tokens. Optional fast mode is $16 / $80 and can be up to 2.5x faster. Anthropic also says Fable 5.2 and Mythos 5.2 generate outputs up to 30% faster than the 5.1 generation.

6. Read the benchmarks as "when, not always"

On FrontierCode 2.0, Fable 5.2 reaches 49.1% at Xhigh effort. At High effort it roughly matches Opus 5.5 at about half the cost per task. Other official scores include FrontierSWE v2 at 69.9% and OSWorld 2.1 at 89.3%. Use those as signals for hard agentic coding and computer-use jobs, not as a reason to burn Fable on every autocomplete.

7. Availability this week

Fable 5.2 is available on all platforms, including AWS, Google Cloud and Microsoft Azure. Developers can call claude-fable-5-2 on the API. Anthropic is giving eligible paid plans extra Fable usage in the launch week. Confirm current limits on Anthropic's site before planning production spend.

8. Faceless explainer videos are Greg Isenberg's idea, not our ranking

A YouTube video by Greg Isenberg (Startup Ideas podcast) on ways to use Fable 5.2 (https://www.youtube.com/watch?v=G3i2k6DdI0o) describes one-prompt explainer videos and vertical channels as a business. That is Greg's claim, not an ASC verdict and not something Anthropic's launch post ranks as a use case. If you try the workflow, use Fable to draft scripts from sources you provide, and have a human check facts before you publish.

Keep going

Browse our picks for the AI tools that build the site in the first place: Best for Vibe Coding · Best AI Developer Tools · Best for Research & Writing.

FAQ

Should I replace Opus 5.5 with Fable 5.2 for everything?

No. Anthropic recommends Opus 5.5 as the strongest daily driver for most coding and knowledge work, and Fable or Mythos for the most ambitious open-ended tasks. Keep Sonnet 5.5 and Haiku 5.5 for lighter jobs.

What is the difference between Fable and Mythos?

They are the same underlying model. Fable adds cyber and biology safeguards for general availability. Mythos is available to vetted organizations that need those dual-use capabilities through Anthropic's verification programs.

How much does Claude Fable 5.2 cost?

On the API, Anthropic lists $8 per million input tokens and $40 per million output tokens. Fast mode is $16 per million input and $80 per million output, with up to 2.5x faster generation.

Where can I try it?

Fable 5.2 is on all Claude platforms, with the API model id claude-fable-5-2. Eligible paid plans get extra Fable usage in the launch week. Mythos access is through Anthropic's verification programs.

Sources

Join "The Crunch" Newsletter

Every new AI tool and model worth your time, with a verdict. The Crunch lands every Tuesday.

Prefer audio or short video? Listen to The Crunch Podcast or watch 60-second breakdowns on AI Stack Crunch (YouTube).